Arkansas Tech University The Culture Wars & Political Polarization in Perspective

The OLS models in Table 10.8 regress the trend in group polarization for each of the categories and for religiosity on time. The results are consistent with the religious/secular partisan polarization evident in the group polarization scores in Table 10.7. Each category of religiosity has significantly contributed to polarization on party identification. The regression coefficients for each of the categorical models are significant at the .01 level and on average all of the models explain about half of the variation in religiosity group polarization on party identification. Furthermore, the most or second-most significant contributor to the religiosity polarization on party ID—both in terms of model fit (.537) and the size of the coefficient (0.118)—are the seculars (those who never attend church). While the “Twice a Month” model beats the seculars in overall explanatory value, the coefficient for the seculars is more substantively significant. Again, note that the two categories that are the most substantial contributors to group polarization are the two extreme categories. However the seculars (0.118) out ‘polarize’ the religiously religious by about two fold (0.061). This very much conforms to what one would expect of a ‘culture war’ between the religious and secular members of society.

component of the trend is a significant shift in the average partisan identification of seculars towards the Democratic Party. In other words, seculars are responsible for a significant portion of the religiosity partisan polarization.

On the other side of the religiosity dimension, the regular church attendees have contributed less and less to partisan polarization since the 1970, though they remain the second largest contributor to partisan polarization of the religiosity groups (second to the seculars). The peak contribution of the ‘every week’ church attendees was in the first year of the time series, 1970 (63.43%) in the weighted measure. The second lowest level of contribution to partisan polarization for this group is in the last

year of the time series, 2004 (23.12%). It trails the global minimum in 1990 by only about half a percent (22.26%). From 1986 on, the percent contribution of the regular church attendees falls below the mean contribution for the full time-series. This is less a function of declining polarization among the very religious and more a function of the growing contribution of the seculars.

The secular-religious trends in weighted group polarization on partisanship are illustrated in Figure 10.7. While there is a great deal of year-to-year variation for both the religious and secular groups, the overall trends noted in Tables 10.13 and 10.14 are evident. The “very religious” (i.e. every week church attendees) are big contributors to partisan polarization for the religiosity groups in the early part of the time series, while the seculars take over in the 1990’s and by 1996 are dominating the very religious group in partisan polarization. There is an interesting action-reaction relationship between secular and religious polarization evident in 1988 (Figure 10.7). As mentioned earlier, the 1990’s witnessed a remarkable spike in the number of secular-identifying citizens in the United States. In 1990, the seculars reach their peak in contribution to partisan polarization. This is followed by a ‘reaction’ peak contribution by the very religious in 1994. Which itself gave way to the secular domination of partisan polarization from 1998 forward.

Table 10.15: Weighted & Unweighted Mean Dev. Trend Regressions of Religiosity GP on PID, 1970-2004

Model: GPMD(pd) = B0 + B1(year) + e	Intercept (R.S.E.)	YEAR P.E. (R.S.E.)		R2	N
WEIGHTED GROUP POLARIZATION
Every Week	1314.372 (565.597)	-0.661 (0.284)	**	.229	18
Twice a Month	212.029 (248.855)	-0.107 (0.125)		.028	18
Twice a Year	1173.078 (420.272)	-0.590 (0.211)	***	.282	18
Never	-2699.192 (387.778)	1.358 (0.195)	***	.590	18
Culture War (EW – NEV)	-5398.672 (775.517)	2.717 (0.390)	***	.590	18
UNWEIGHTED GROUP POLARIZATION
Every Week	837.020 (425.461)	-0.421 (0.214)	*	.180	18
Twice a Month	574.398 (449.831)	-0.289 (0.226)		.084	18
Twice a Year	358.608 (291.469)	-0.178 (0.146)		.070	18
Never	-1767.007 (422.639)	0.889 (0.213)	***	.425	18
Culture War (EW-NEV)	-3532.033 (845.840)	1.778 (0.426)	***	.424	18

Returning our attention to Tables 10.13 and 10.14, the middle religiosity categories show no increasing partisan polarization. Both of the middle categories show declining contributions in the weighted measure—registering below-10% contributions on multiple years after 1988. The unweighted contributions, which are fairly stable over the time series suggests this decline is a consequence of fewer citizens identifying as semi-regular or even occasional church attendees. The maximum percent contribution to the weighted partisan polarization measure for the month church attendees was in 1988 (49.67%). The largest contribution on weighted group polarization for the ‘once or twice a year’ church attendees group was in the prior survey year, 1986 (45.38%). Both minimums come in at 1990 (7.35%) or later (7.23% in 2004).

The percent contribution regression models for religiosity on partisanship are reported in Table 10.15 for both weighted and unweighted group polarization. For the weighted models, the two most substantively significant trends are the declining contribution to partisan polarization by the ‘every week’ church attendees (-0.661) and the increasing contribution of the seculars who ‘never’ attend church (1.358). The contribution of the very religious to partisan polarization has declined by over half a percent for every survey year, while the secular contribution has risen by 1.4% in every survey year of the time series. The linear trend model for seculars explains nearly 60% of the variation in mean deviation of percent contribution to partisan polarization (R² = .590). The “Culture War” model tracks the absolute difference between the percent contribution of the every week church attendees and the secular citizens (mean deviation). The difference in contribution between the very religious and secular respondents increased nearly 3% for every survey year (2.717), and this model also accounts for close to 60% of the model error (R² = .590). The one other significant trend is the declining contribution of the “twice a year” church attendees (-0.590), with it declining over half a percent a survey year since 1970. The reduction in model error for the ‘infrequent’ church attendees was nearly 30% over the past 34 years (R² = .282). This appears to be exclusively a function of the declining number of citizens who report going infrequently, as the unweighted model for this category is not statistically significant (R² = .070). Likely these are citizens who were weakly attached to their church or religion and, over the course of the past thirty-plus years, have moved into the seculars and now never attend church.

In the unweighted group polarization models (weights each category of religiosity equally), the significant trends identified in the weighted models (excepting the “twice a year” model) are also significant, though the coefficients are uniformly lower than in the weighted models. Clearly the decline of religiosity accentuated the divergence of the religiosity groups on party identification. The very religious group (every week) decreasingly contributes to partisan polarization (-0.421) and the secular group increasingly contributes to the group polarization measure on partisanship (0.889). The “Culture War” model indicates that the divergence of religious and secular citizens has increased just short of two percent per survey year irrespective of the change in the size of the religiosity groups (1.778). This model explains about 40% of the variation in the group polarization measure on partisanship (R² = .424).

Religiosity Polarization on Ideology

The significant group polarization of religiosity on partisanship apparent in the previous analysis is echoed in the significant group polarization of religiosity in the ideological dimension. Table 10.16 has the group polarization measure for the religiosity groups on ideology. Again it is apparent that significant polarization has occurred over the time series. All four categories have witnessed substantial increases in their contributions to group polarization since the 1970’s, with once again the most significant contributor being that of the seculars. Note that in 1972 the secular contribution to polarization on ideology is half that of the every-week church attendees. If you will recall, it is not merely the relative distance on the dimension which determines a group’s contribution to polarization but also the size of the group relative to the population. With seculars on the rise over the course of the time series, they increasingly contributed to the ideological polarization of the religiosity groups.

Table 10.17 reports the trend models for the religiosity groups on ideology. While the secular model does not provide the best proportionate reduction in error of the categorical models (that honor belongs to the every-day-a-weekers), substantively speaking the seculars have seen the largest increase in their contribution to ideological polarization (.177 for each survey year). The middle two categories contribute relatively little to the trends in group polarization on ideology. Both the “Twice a Month” category (0.029) and the “Twice a Year” category (0.021) have relatively small coefficients. Their impact combined is only half that of the regular, weekly church attendees (0.093) and not even in the same ballpark as the seculars. That said, while their substantive impact is negligible, they both have a positive effect on religiosity group polarization on ideology and the model fit statistic indicates that both models improve on the mean by an over 50 percent reduction in error. For the full religiosity classification, there

is a 0.319 (four tenths of a point) increase in polarization for each survey year for the regression model. The R-Square for this model (.549) indicates a good fit and thus yet another example of increasing political polarization consistent with the culture wars thesis.

The percent contributions of religiosity to group polarization for the weighted and unweighted measures on the ideological dimension are reported in Tables 10.18 and 10.19. The decline in the contribution of the most frequent church attendees of the religiosity groups isn’t as steep in the ideological dimension as it was on partisanship, but it is significant and pronounced. There are higher percent contributions in the weighted measure than we saw with partisanship, as the “every week” category accounts for at least 60% of the group polarization on ideology in 1992 and 1994. The two middle categories trail behind the two extreme categories significantly in their contribution to weighted polarization on ideology. The maximum contribution for the “twice a month” category was 10.3% in 1984 and, for the most part, the contribution of this category lags well below 10% for the time series. While the “twice a year” group had a spike in its contribution in 1996 (24.02%), this was the only time since 1988 that their contribution to polarization exceeded 10%.

The unweighted contribution to group polarization for the “every week” group tracks with the weighted contribution, however it is between 5 and 15% less of a contribution in the unweighted measure. The size of the group clearly accentuated its contribution in the 1970’s and 1980’s and its subsequent decline downgrades its contribution in the 2000’s. The secular group exhibits the same steady trend towards greater contribution to the polarization of religiosity groups on the ideological dimension. It contributes nearly half of the polarization for each of the survey years in the first decade of the 21^st century (47.26% - 47.77%). Again we see the impact of group size on the secular contribution to polarization. In the 1970’s its weighted percent contribution lags behind its unweighted percent contribution. In 1972 seculars contributed 35.29% of the unweighted polarization on ideology, while it only contributed 24.91% in unweighted ideology. This relationship flips in 1990. In 2004, the unweighted group polarization was 47.77%, but the weighted group polarization was higher: 56.29%.

The percent contributions for both weighted and unweighted group polarization of the religiosity groups on ideology are given in Table 10.20. While the middle categories exhibited no significant trends in their percent contributions in the partisanship model with one exception, both middle categories exhibit significant contribution trends in the weighted models. The “twice a month” group has a statistically significant increase in its contribution (0.097), while the “twice a year” group had a declining part in group polarization over the time series (-0.311). The trend for the group that reports attending church a few times a month accounts for more of the variation in the group polarization measure (R² = .402) than does the twice a year model (R² = .270). Both significant trends are a consequence in the variation of group size, as when each category is not weighted by their percentage in the population (as approximated by sample size), the linear trends are not significant.

Using the group polarization measure, I have examined polarization on two different dimensions (partisanship and ideology) for two distinct groups (class and religiosity). Despite the colorful story-telling of Thomas Franks, I find little evidence of income group polarization. And what polarization exists operates more at the ‘rich’ end of the income spectrum rather than in the poor and working class groups as suggested by Franks. Lastly, I find significant polarization among the religiosity groups with a significant proportion of the partisan and ideological polarization among religiosity groups attributable to the rise of the seculars in society over the last three decades. It is along the religious divide that America has become increasingly divided ideologically and along which the parties have increasingly divided themselves in electoral and policy competition. While there is slight polarization of the classes, it is neither the poor nor the working class that have been the prime movers for polarization in partisanship or ideology. Hunter’s original insight was correct. The growing culture wars have a distinct religious dimension with partisans and ideologues increasingly divided based on their belief in God and affiliation with religious institutions and organizations.